{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
excited ???
it A FAKE i mounted the quadrant data folder as a tmpfs meaning ramdisk mount
Very old thread.. but some things not resolved here really as far as I can search.
first
1) Yes this is fake, because there is no "disk" writing ever going to happen. The data will be lost. Everyone agrees.
2) However, if you used an fsync violating write cache.. like loopback you can get a similar score (we all know this too) but that's NOT fake (as many people say), that's just saying that write caching improves short burst intermittant write performance which is ABSOLUTELY TRUE (and can remove lags!) but it isn't a measure of I/O throughput AND since many types of realistic writes may already be cacheing one way or another (not fsyncing), it may or may not be that relevant.
Now the problem is the logic goes that quadrant gives this "fake" score ergo quadrant is useless ergo epics low score on stock is meaningless and especially since we all know epic doesn't lag.
Problem. epic doesn't lag because it's writes are supposed to be more aproximately like zero time than like SGS time (say 1/4 SGS time at worst to make up a fake number). That means no matter what you do, force fsyncs, avoid caching etc... the score should be pretty high (closer to 2400 than to 900 one would think).
None of this explains why epic gets a low score on unmodified stock roms. I'd expect 1500 at the very least. Something is still missing in the understanding here. Yes I know, one should look at full quadrant, ok but that argument implies something else in the epic is holding the score back alot edit: and that theory doesn't hold up either because then the "fixed" score should not be as high as it is. It's funny that both the low and high scores are almost identical to the SGS.
The answer may still be quadrant is garbage, but it would have to be garbage in a new way, not just in that it gets fooled by ram disk way (edit: which isn't really its fault anyway, although if it wrote a larger file it might not get "fooled" by a cache, but that argument depends on what you want to measure).
appagom said:
Very old thread.. but some things not resolved here really as far as I can search.
first
1) Yes this is fake, because there is no "disk" writing ever going to happen. The data will be lost. Everyone agrees.
2) However, if you used an fsync violating write cache.. like loopback you can get a similar score (we all know this too) but that's NOT fake (as many people say), that's just saying that write caching improves short burst intermittant write performance which is ABSOLUTELY TRUE (and can remove lags!) but it isn't a measure of I/O throughput AND since many types of realistic writes may already be cacheing one way or another (not fsyncing), it may or may not be that relevant.
Now the problem is the logic goes that quadrant gives this "fake" score ergo quadrant is useless ergo epics low score on stock is meaningless and especially since we all know epic doesn't lag.
Problem. epic doesn't lag because it's writes are supposed to be more aproximately like zero time than like SGS time (say 1/4 SGS time at worst to make up a fake number). That means no matter what you do, force fsyncs, avoid caching etc... the score should be pretty high (closer to 2400 than to 900 one would think).
None of this explains why epic gets a low score on unmodified stock roms. I'd expect 1500 at the very least. Something is still missing in the understanding here. Yes I know, one should look at full quadrant, ok but that argument implies something else in the epic is holding the score back alot. Is that true?
The answer may still be quadrant is garbage, but it would have to be garbage in a new way, not just in that it gets fooled by ram disk way (edit: which isn't really its fault anyway, although if it wrote a larger file it might not get "fooled" by a cache, but that argument depends on what you want to measure).
Click to expand...
Click to collapse
I'll explain it right now. Benchmarks are rarely an all-encompassing number that can describe the overall performance of a device. It is impossible in a situation like this.
There.
That's all.
At least this isn't like the video card industry, where products are designed specifically to perform well in benchmarks, which pleases all the drooling fans but ignores the actual goal of real-life performance.
Then there's the bribery and tricks on both ends to get specific products to do better or worse.
Please don't help get us there. Ignore this stupid benchmark.
I understand this fully... which you would know if you read my post. That doesn't answer the question though.
As for the ram cheating(a separte issue from my question really) here's a question... if you have 5GB internal RAM cache attached to a painfully slow block device is that cheating? What if it has an internal battery backup so it can take all night to catch up. Is THAT cheating... On a high load server, of course it is. On a cell phone.. no way that's cheating, and in fact there are rumors it's in development (well maybe not 5GB).
Why thread necro? I posted a thread about this just the other day with a 2597 score.
[2597] This is why friends don't let friends use Quadrant.
fine.. I'll read there (think I've seen it, but maybe wasn't thinking about this at the time but thought it didn't address this) and repost there in a couple of day if it's not answered there already. I guess you can consider this one informally closed for the moment. I don't have time today to find out actually.
Edit: BTW the reason I ask is not cause I care if quadrant describes epics performance well (if it runs smooth it runs smooth of course). The reason I ask is because I'm curious about the lag fixes on other devices and understanding this might provide some useful insight or overlook.
Related
Hey guys,
I've been doing these benchmarks alot on WinMo before I got my first Android device. I always felt Winmo was sluggish, but look at this:
GLBenchmark link
I compared the Nexus with a Milestone, HD2 and 3GS. And it shocked me... The HD2 is faster at most benchmarks!
Is that because of the SDK with OpenGL ES 2.0 isn't used yet?
Hoping this could be improved, because I left Winmo because of this.
I don't believe this benchmark.
Okay, you're comparing several different platforms here. The HD2 and the Nexus One have the exact same GPU.
More accurate: this particular benchmark is faster on its Windows Mobile port than on its Android port.
scoring 100% in a benchmark can not be right?
frleus said:
scoring 100% in a benchmark can not be right?
Click to expand...
Click to collapse
Generally speaking in these types of benchmarks, the benchmark is set by the highest-scoring device. 100% would be the king for that specific test.
You would be correct, Android 2.1 as it is on our phones doesn't have all of the graphics boosters. To bo honest, it isn't that surprising IMHO, since these benchmarks are hugely influenced by the software itself, otherwise ROM cooking wouldn't do anything for increasing these #'s.
Maybe you could re-run this with a Cyan ROM, which has lots of those little things added in, such as the Arm Neon graphics inhancing(or whatever the heck its exact name is).
Also, keep in mind, that there are more important factors than "benchmarks".
pjcforpres said:
You would be correct, Android 2.1 as it is on our phones doesn't have all of the graphics boosters. To bo honest, it isn't that surprising IMHO, since these benchmarks are hugely influenced by the software itself, otherwise ROM cooking wouldn't do anything for increasing these #'s.
Maybe you could re-run this with a Cyan ROM, which has lots of those little things added in, such as the Arm Neon graphics inhancing(or whatever the heck its exact name is).
Click to expand...
Click to collapse
I agree, with this, but adding one more thing instead of just testing different phones, test these benchmarks on cyan and emnother. With the nexus our graphics performance is aided by software so hardware isnt going to be the kicker between the hd2 and the nexus.
CM 5.0.6
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Web based benchmarks are as fail as the DPI on the HD2.
TheDudeOfLife said:
Web based benchmarks are as fail as the DPI on the HD2.
Click to expand...
Click to collapse
In your experience which would be a good benchmark besides Linpack?
jlevy73 said:
In your experience which would be a good benchmark besides Linpack?
Click to expand...
Click to collapse
Water submersion test.
TheDudeOfLife said:
Water submersion test.
Click to expand...
Click to collapse
+ Drop Height.
uberingram said:
+ Drop Height.
Click to expand...
Click to collapse
Submersion test + drop height = the angle of the dangle
I agree that static benchmarks aren't anything, but the Nexus should be at least able to score comparable to the HD2, right? Hardware is the same, it's just the software that differs... alot.
NeoS2007 said:
I agree that static benchmarks aren't anything, but the Nexus should be at least able to score comparable to the HD2, right? Hardware is the same, it's just the software that differs... alot.
Click to expand...
Click to collapse
ah the quest for improving graphics on msm7k errr i mean snapdragon processors begin. Nice to see you've jumped to Android neoS2007
i think theres heaps of room for n1 to improve and count me if you decide on creating a group for improving graphics on the n1
babijoee said:
ah the quest for improving graphics on msm7k errr i mean snapdragon processors begin. Nice to see you've jumped to Android neoS2007
i think theres heaps of room for n1 to improve and count me if you decide on creating a group for improving graphics on the n1
Click to expand...
Click to collapse
Good to see you here too babijoee. I'm trying to get someone on the MSM bandwagon to coordinate it, but Snapdragon is just so much more interesting... Anyways, goodtimes.
Anyways, I'm not setting up a group until i'm fairly certain the N1 CAN get better. I want to compare the N1 with the Desire and then again when OGLES 2.0 is fully implemented.
Nothing official but defiantly interesting:
http://www.androidspin.com/2010/04/21/google-getting-ready-to-dish-out-some-froyo/
I'm sure if everything mentioned here would defiantly give the n1 a substantial performance increase
Definitely worth waiting for. I'll see if I can track improvements in the graphics area of the N1 once Froyo is out.
NeoS2007 said:
Good to see you here too babijoee. I'm trying to get someone on the MSM bandwagon to coordinate it, but Snapdragon is just so much more interesting... Anyways, goodtimes.
Anyways, I'm not setting up a group until i'm fairly certain the N1 CAN get better. I want to compare the N1 with the Desire and then again when OGLES 2.0 is fully implemented.
Click to expand...
Click to collapse
I just ran it on my Desire. The upload failed so i dont have exact figures to prove anything, but it looks much the same as the N1 listed on their site. Most things are within 5%. My HD2 running a dutty custom ROM and Chainfires driver hack is a bit faster.
Raging Thunder 2 run extremly fast on N1, faster than moto chaser on ipod touch 2g. Anyway mobile benchmarks are always bulls**t
This might not be the right fourm but could someone please tell me why HTC phones are twice as fast than the Epic on linpack? I've tried a EVO and a Droid Incredible, both get a score of 32.500! Why?
Because JIT is optimized for Snapdragon processors, you see more performance boosts with those phones. However, mere benchmarks do not translate to real world performance whatsoever. My Epic is faster than any phone that's out there. Never go by benchmarks, lot of them are as useless as the Windows Experience Index crap. Our phones have the better hardware, when official Froyo is out and devs get the source, you'll see how powerful it really is.
Because linpack performance does not equal real life performance.... I also recall reading that linpack has not been optimized for the epic's processor. However, try 30 seconds using an evo and compare it to the epic and you'll have your actual answer
Sent from my SPH-D700 using XDA App
I think they are a 128 bit system versus 64 bit. But all I hear from evo user is the epic is faster
Sent from my SPH-D700 using Tapatalk
nimaim said:
Because JMy Epic is faster than any phone that's out there.
Click to expand...
Click to collapse
Just like cars, there's always someone faster...I think it is pure ignorance to state you alone have the fastest phone in the whole world. *shrug* For me my user experience is awesome and i know firsthand that benchmarks don't tell the whole story and can be faked. With tmpfs and a little knowhow anyone can 3000+ on quadrant.
Sent on the go from my mobile device
I didn't mean to come across as arrogant. I didn't mean I alone have the fastest phone in the world at all. I meant with everything that's out in terms of OC'd kernels, Epic owners in general have faster phones than practically any other out there (currently). Benchmarks are not high, but that does not translate to real world performance which is what I was getting at.
Essentially what has been said before: Linpack has been compiled and optimized for the other processor. Someone has to write code for each specific processor in order for it to take advantage of each one's architecture. Back when Intel's Nahalem chips came out, similar results were observed (I used to build HPC clusters).
Try nenamark1. The epic smokes the evo in that test
Sent from my SPH-D700 using Tapatalk
Simple, linpack is a test checking for raw CPU power via floating points..Snapdragon has a thing called a VFP which is a Floating Point Accelerator...
Aka instead of checking raw CPU power of the CPU..your checking how fast the the VFP processes floating points lol..
To give an example, its like one person actually taking a test, while the other is copying off an answer sheet..who will be the fastest?
If you go by benchmarks that have not had their code optimized for the Snapdragon, where it is more apples to apples (it will never be completely apples to apples), the Hummingbirds generally come out on top.
Plus, if you look at gaming, the FPS on the Hummingbird is much faster. Of course, CPU and graphics benchmarks are quite different.
It has a keyboard
Thread moved to General.
I'm pretty new to android phones but I did my homework before i decided on my epic. My wife wanted the Evo for the larger MP camera and bigger screen so we ended up with one of each. Form the time i've spent on my wife's evo and the time she has spent with my Epic we both have to agree that the Epic preforms far better in realworld test. The scrolling is smother and it has a better overall experience than the evo. So forget about benchmarks... just try one head to head and you'll see which one is the better device..
I've said this before and will say it again. My rooted evo with cm6 and oced to 1.12 ghz isn't as smooth as my epic was out the box. My evo gets 1500 on quadrent while my epic gets 1100 with no jit. Now you tell me wich phone is faster?
Sent from my SPH-D700 using XDA App
gTen said:
To give an example, its like one person actually taking a test, while the other is copying off an answer sheet..who will be the fastest?
Click to expand...
Click to collapse
The person taking the test will be faster since the copier has to wait for the answers.
Facepalm...
Sent from my SPH-D700 using XDA App
crabjoe said:
The person taking the test will be faster since the copier has to wait for the answers.
Click to expand...
Click to collapse
...your assuming he is copying from the other persons answer sheet and not the teacher's answer sheet...must we really get into this? >.> lol
Either way thats the reason...
crabjoe said:
The person taking the test will be faster since the copier has to wait for the answers.
Click to expand...
Click to collapse
Winner by TKO.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Squirel aids!
Looks like google got JIT situated on hummingbird CPU
sorry crappy screen haven't unlocked boot loader yet
Now off to unlock to help out with system dump
So dev can try to port
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Sent from my Nexus S using XDA App
This is awesome news as long as we can get GB ported
What does the blue portion of the bar represent?
blanked said:
What does the blue portion of the bar represent?
Click to expand...
Click to collapse
The CPU. That's what was lagging with the current froyo builds. The JIT compiler was optimized for QCOM chips, but not the Hummingbird processor. Looks like that's been sorted out.
Is that available only in the pro version (I don't get the colors just one solid bar)? What is green and how come Droix X kicks so much butt with it? Memory?
g1ddy said:
Is that available only in the pro version? What is green and how come Droix X kicks so much butt with it? Memory?
Click to expand...
Click to collapse
Blue CPU/Red Memory/Green I/O/Grayish Yellow 2D/Yellow 3D
SO WANT those speeds...
Considering we're running the same processor...can't wait!
Awesome. Dump!Dump!
Don't blame me, blame my keyboard's autocorrection algorithm.
This was linked to the Vibrant section, and noticed the lower (smaller block) on 3D performance? Can anyone confirm?
edit: damn thought this was the Nexus S section, lol.
RobBull69 said:
The CPU. That's what was lagging with the current froyo builds. The JIT compiler was optimized for QCOM chips, but not the Hummingbird processor. Looks like that's been sorted out.
Click to expand...
Click to collapse
the test breakdown is available in Quadrant Advanced. you can get it in the SlideME market.
and the green represent I/O which doesnt really contribute to a lot of performance... you will notice that people running OCLF will have quadrants score of 2k+ mainly due to i/o exploit
what's the linpack on it?
THIS IS EXACTLY what I have been waiting for! I have been one of the most vocal people about the Dalvik Jit Compiler not offering the Hummingbird the same performance gain that scorpions received, and it seems like Google definitely sorted this out. Totally just put new life into the Hummingbird. Google choosing the Samsung Galaxy S as being their new Nexus phone was the best thing that happened to Vibrant owners.
Thats just a little higher than my Vibrant with Voodoo Lagfix. Which means thats its either 'meh', or if its getting the result without being I/O lagfixed, then... 'WOW!'
Mannymal said:
Thats just a little higher than my Vibrant with Voodoo Lagfix. Which means thats its either 'meh', or if its getting the result without being I/O lagfixed, then... 'WOW!'
Click to expand...
Click to collapse
The Nexus S doesn't use RFS, so there's no need for a lagfix in the same sense as the Galaxy S phones. However, Google did neglect to set one of the EXT4 mount options, so there is a Voodoo kernel now that shows significant improvement. They're getting around 2400 with the new kernel. http://forum.xda-developers.com/showthread.php?t=876132
you can get the same thing on a SGS with the CM6.1beta
I played with one for a short time at Best Buy tonight. It is ridiculously fast. It loaded Angry Birds loaded and ready to play in the same time it took my phone (running Eugene's Ginger Clone 2.2 with the super I/O kernel) to bring up the loading screen. I understand the browser still has some of the same lag problems, but I didn't get to play with it long enough to see for myself.
If devs can port the 2.3 kernel to the Vibrant, then I'll stick with it because it will indeed gain what this phone can do, but If now then I think I'll get the Nexus S. The fact that Google took Samsung's hardware and made it sing like this and Samsung can't even manage a workable 2.2 yet has me seriously contemplating jumping ship. I don't think I'll sell the Vibrant though, the community and developers have been great and I want to see exactly how far they can push the phone, but it'd be nice to daily drive something you know is fully supported.
Mannymal said:
Thats just a little higher than my Vibrant with Voodoo Lagfix. Which means thats its either 'meh', or if its getting the result without being I/O lagfixed, then... 'WOW!'
Click to expand...
Click to collapse
hey mannymal, I think you fail to see the point and significance of this. The total end score isn't what we are excited about, at least not me. The cpu score is what we are excited about, and you can only see the score breakdown in Quadrant Advanced or Quadrant Professional.
The dalvik jit compiler in froyo 2.2 was not giving the hummingbird the same CPU performance gain that was being achieved by the scorpion chips. Meaning that a crummy old first generation scorpion was outperforming the hummingbird in cpu tasks.
The significance of this thread and the quadrant photo, shows that Google appears to have hopefully sorted things out, and that the Hummingbird is now receiving a CPU performance gain that was being partially neglected in 2.2
I hope you realize the end score wasn't the focus, and that no matter how many lagfixes you apply, your cpu performance doesn't necessarily go up.
Is that all? Very low score! I was expecting better than this for a Nexus S ..
scrizz said:
you can get the same thing on a SGS with the CM6.1beta
Click to expand...
Click to collapse
But sure, I get the same score with my SGS Quadrant XXJPY and LagFix!
Hey guys I was just wondering what's your best Quadrant score you've had on your HD2? I got what I think is a fairly high quadrant score today and I was fairly impressed.
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
Sent from my HTC HD2 using XDA App
i think 2300 or 2700 somewhere in between
I don't know the reasoning behind it, but from what lots of others have said, Quadrant scores are totally irrelevant. I can only assume it's because we're using ported software, rather than something that was built specifically for our devices. That's my guess anyway.
You beat my score!
johncmolyneux said:
I don't know the reasoning behind it, but from what lots of others have said, Quadrant scores are totally irrelevant. I can only assume it's because we're using ported software, rather than something that was built specifically for our devices. That's my guess anyway.
Click to expand...
Click to collapse
Not totally irrelevant, but sometimes close to it. Quadrant standard is useless, pro is a little better.
The problem with Quadrant is:
1) Disk IO test, if you try some of the builds with very high score you'll see it's coming from IO, you can just mount that one part that Quadrant use in RAM, get crazy high score, but that won't speed up any other apps/system as you still use your NAND/SD card
2) Video speed, as on early Darkstone builds video score was very low as that build had something turned off (because with it's ON there were other bugs), but all apps were using some other API for video, so only Quadrant was slow on it
3) Quadrant tests doesn't show real system performance, so higher score might have no increase in speed in other apps. For example you can put newest CPU on your PC, but having only 256mb RAM would make everything slow, no matter how fast your CPU is, but CPU score would be higher
This doesn't mean that Quadrant is bad, in my opinion if you're using pro version and you get some component score (for example CPU score) higher on one build then on other - that's good. If you get total score higher, build might be actually slower then other build with low score.
this is the old 1.32... at 1200mhz cedes sd build.
with the new 1.72... dhd base i get max 5000 cpu points and approx 5000 I/O points with data on ext4 partition. that makes 2400 overall quadrant points. at 998 mhz. but if i put all data to nand the I/O points are 1250, but the device feels faster and boots faster. for now i only put app folder on the ext4 partition, dalvik and data remain on internal memory. i am still testing whats the best combo, it takes lot of time
johncmolyneux said:
I don't know the reasoning behind it, but from what lots of others have said, Quadrant scores are totally irrelevant. I can only assume it's because we're using ported software, rather than something that was built specifically for our devices. That's my guess anyway.
Click to expand...
Click to collapse
I think your right there.
I don't use Quadrant much but i got over 3000 before (using Darkstones SUPERAM)
Mine was when I didn't even overclock the CPU.
Thought that was pretty good, I usually get about 1600.
Sent from my HTC HD2 using XDA App
Skellyyy said:
Mine was when I didn't even overclock the CPU.
Thought that was pretty good, I usually get about 1600.
Sent from my HTC HD2 using XDA App
Click to expand...
Click to collapse
thats not bad man what build do you have on your phone?
SuperNAND got about 4000.
Proof:
Someone beat this . Quadrant scores are garbage, it proves nothing.
jonny68 said:
thats not bad man what build do you have on your phone?
Click to expand...
Click to collapse
As I'm on my phone right now I can't get the link but its the DHD build by Andrew something. Should be on the first page in the NAND forum.
Sent from my HTC HD2 using XDA App
I've just saw Galaxy S Antutu Benchmark screens that had 3900 score @ ~~1400MHz and I'm wondering why our device even if it's pretty similiar does score only 2900 ? It's a huge difference.
Even oldie like tuned Xperia X8 can score 2400 which isn't even ARMv7 and has 256MB RAM with older GPU.
I'm a bit disappointed by our ROMs, because there are many of them but none is perfect. ILWT seemed to me the best (and it is), but has lower score than ARHD, which needs swap and many improvements to work smoothly and it isn't still capable running games like GTA3, MC3, ICS can't run them too because phone just reboots (ICS is in beta stage so I'm not looking at stability yet, but that's bad ).
ILWT result in best performance as for GTA3, this game was running great.
Here's screen of SGS Antutu:
http://desmond.imageshack.us/Himg855/scaled.php?server=855&filename=1331989260894.jpg&res=medium
X8 (for reference, score is without sdcard): http://iv.pl/images/61942066283720320721.png
and our Desire Z's in this thread, which is highest
http://forum.xda-developers.com/showthread.php?t=1560772
Why devs won't focus on improving performance? schedulers, more oc values, governors isn't everything, why then SGS can get 2x more score in RAM if it has 512MB too. Overclocking in SGS by 100MHz steps improve score by ~200, and in Desire Z it looks like it's not working as good because I didn't noticed that BIG boost, just ~50pts. I know it's not that easy to develop, but we're alot behind with performance than SGS.
SGX540 is probably better GPU and there's other processor, but if stock software had 3000pts in SGS, and they(devs) managed it to improve by 1000pts our had 2600 and got improved by 300 only.
I try to be optimistic, as I respect all the work done by developers. Every software IMO is good for everyday usage, but well, by modding I though it could be bit better
I am just dropping my two cents here, but I think that the devs go by the thought that benchmarks are - as Microsoft put it - not relevant to real day-to-day use scenarios. A phone with awesome benchmarks (or a ROM such as ARHD) may act quite sluggishly and crappily in day-to-day usage without a ton of tweaks and stuff (like ARHD does).
So, you see... the mechanic tells you that you can get 300km/h in your new Toyota, but at 160km/h the wind is screeching in the inside of the car so loud that you would *not* want it to go to 300/h.
I hope you understand my analogy. The point is - take benchmark results with a grain of salt.
I know what you mean, also saw people talking about desire z being smoother than sgs in stock config, but also there's something that sgs devs did if scores went over 25% higher
As far as I know:
benchmark =! reality
And the desire Z is worse than sgs, the gpu is worse, don't know about cpu, the ram may be different not only from how much it is (even if I've seen they're both ddr1)
Apart from this I'm not saying that we'd better have not more than 10 roms and more devs trying to add something very new to the phone. We've many ICS roms, many cyano roms, many sense 3.5 roms, many ics sense roms, many miui roms. I'd rather have two of each group and ie usb host or something a bit trickier... I mean, with all the respect possible for developers of every kind, hacking is becoming not hacking anymore, I mean everyone can with a bit of knowledge make his own rom (I remember of people under fifteen...). But hacking, imo, is when siul, OdienManSam, taz were trying to port ICS, all the work that many here are doing making things work in ics (Flinny, blk_jack), what on One X forums people are trying to do to gain S-OFF, when OdienManSam was trying to port usb host (will never know if he succeded)... I think there's much more juice to squeeze from our phones, less people willing to do it...
I'm not deplorating devs works, just my opinion...
If it means anything:
{
"lightbox_close": "Close",
"lightbox_next": "Next",
"lightbox_previous": "Previous",
"lightbox_error": "The requested content cannot be loaded. Please try again later.",
"lightbox_start_slideshow": "Start slideshow",
"lightbox_stop_slideshow": "Stop slideshow",
"lightbox_full_screen": "Full screen",
"lightbox_thumbnails": "Thumbnails",
"lightbox_download": "Download",
"lightbox_share": "Share",
"lightbox_zoom": "Zoom",
"lightbox_new_window": "New window",
"lightbox_toggle_sidebar": "Toggle sidebar"
}
I'd say that it's pretty close to what the SGS performed in the benchmark (considering I was actually playing music whilst testing). Regardless, benchmarks are indeed far from everyday use scenarios.
Andromadus @ 1.2GHz. As the dude above me states, our device really doesn't score that much lower all things considered.
Sent from my HTC Vision using xda premium
Mine is at 3400, too with Andromadus ICS ROM @1,2GHz. And tested up to 1,6GHz running stable. So I have some reserves if I like, but it´s smooth enough at 1,2GHz, so why bother?
Mike